Though next-generation sequencing has been widely adopted in human research, particularly for gene-expression profiling, microarrays still remain a key platform for plant and animal researchers for reasons of cost and throughput, according to several experts familiar with the space.
And while Affymetrix and Illumina have to date debuted fewer than a dozen catalog genotyping chips for agricultural research combined, executives from both firms believe that most plant and animal researchers will eventually adopt SNP genotyping arrays for a number of different applications, and are planning a host of new agriculture-focused product introductions.
"We are seeing a huge amount of studies on everything from cassava to turkey, salmon, aquaculture, animal husbandry — every crop that's out there, you name it, it is already being worked on to generate all the discovery sets by next-gen sequencing," Tristan Orpin, Illumina's vice president and chief commercial officer, told BioArray News at the Plant and Animal Genome Conference held last week in San Diego.
According to Orpin, next-gen sequencing is being used to create "vastly more comprehensive genetic maps" of different plants and animals of interest, and the "next step pretty much universally at this time is to convert that data into a range of different arrays."
The trend of different plant and animal research consortia adopting genotyping arrays for applications such as genomic selection or association studies has been supported by the lower cost of sequencing and has been ongoing for several years (BAN 2/1/2011). These consortia use sequencing to generate data that is later used to inform the design of the arrays.
Arrays have been generated for a number of livestock communities, with bovine research leading the way. Today, the cattle industry has at least five catalog arrays to choose from. Illumina offers its BovineHD, the BovineSNP50, the BovineLD, and the Bovine3K Beadchips, each of which differs in terms of density and content, while Affy offers its Axiom Genome-wide BOS 1 Array. The BovineSNP50 hit the market in 2008, while BovineHD became available in 2010. The BovineLD, Bovine 3K, and BOS 1 array all became available last year.
The cattle industry's head start in seeing arrays developed for its needs can largely be explained by its preparedness to integrate array-based findings into breeding decisions, as well as the high cost of raising the animals, according to Orpin.
"The reason why it's mainly bovine is because the dollar value per animal is so significant and because within cattle breeding associations, the integration of systems supporting the optimization of cows for dairy and meat production is so much more advanced than all the other livestock," he said.
Illumina sells other commercial ag-focused genotyping arrays, including the OvineSNP50, PorcineSNP60, CanineHD, and MaizeSNP50 BeadChips. Meantime, Affy launched its GeneChip Rice 44K SNP Genotyping Array last year and expects to later this year debut a high-density chip for poultry studies (see related story, this issue).
Orpin said he expects that other communities will follow the "same scenario that Illumina had with bovine," in which it moved from a higher- or middle-level-complexity tool that was used primarily "as a set for mapping validation and discovery in those populations" to less complex "screening sets that can be used at a much lower price point — so going from hundreds of thousands of markers to maybe 50,000 markers to 5,000 markers [or] 3,000 markers, the most useful [number of] markers for solving an economic problem in breeding."
Affy has a similar view of the market. Jasmine Gruia-Gray, Affy's vice president of global marketing, told BioArray News that genotyping arrays have “great value in a breeder's toolbox in the genomic selection process for improving yields and quality.” The company earlier this month announced a deal with Chinese genomics institute BGI that could see Affy debut a number of other ag-related SNP arrays as soon as the second quarter (BAN 1/17/2011).
Affy is positioning its arrays for both discovery and validation. The firm's arrays are "just as valuable in plant and animal genomics research for uncovering nucleotide diversity and genetic origins in plant germplasm collections [as they are] in validating SNPs that have been identified through other technologies," said Gruia-Gray. She added that the company's arrays could be used for a number of applications, including diversity analysis, association studies, and genomic selection.
The Appeal of Arrays for Ag-Bio
Though Illumina is clearly keen to see its next-gen sequencing instruments used in SNP discovery — all four of the company's presentations during its workshop at PAG focused on sequencing-based projects — Orpin said that arrays are more likely to be used in coming years for high-volume, agriculture-related studies.
The appeal of arrays is based on economics and the adaptability of array-based assays into field locations, "which means that it is based on the throughput and simplicity of the assay and how those metrics align to the needs" of users, said Orpin.
"In a lot of these areas, the dollars you can spend on animal or crop research is going to be a fraction of what you can spend on human research," he said. "And you need to be able to turn around thousands of samples in a very short time because of the way breeding cycles work," he said. "At the moment, next-gen sequencing does not provide the economic solution or throughput needed to provide this dataset in a screening mode."
Above all, the cost of such studies must be proportional to the investment in the crop or the animal, according to Orpin. Because of this, arrays are a "consistent next step" for researchers, he said. At the same time, he expects this dynamic to change as sequencing costs drop.
"In five years' time, [the market will reach] a point where it makes more sense from a cost perspective and also maybe from a throughput perspective to do genotyping on next-gen sequencing," Orpin said.
Still, he said that data-analysis challenges may slow the adoption of next-gen sequencing in the ag-bio community.
"The limitation of arrays is that you only get what you are looking for, but the lovely thing about arrays is that you only get what you are looking for, so … you only have to analyze what you are expecting, and in these applications you don't want all this additional information — that is just background noise, additional data you have to sift through," said Orpin.
"Even five years from now, I could be completely wrong, it may move largely to next-gen [sequencing], but at the same point you might find that five years from now because of the price point and throughput, a lot of the analysis stays on arrays," he said.
Going forward, both Illumina and Affy have pledged to launch more ag-focused chips. "This year … there will be a whole series of new arrays at different levels of complexity supporting different animals," said Orpin. He declined to further discuss future product launches. He also noted that commercialized arrays are just a small slice of the total number of custom ag-related chips that the company is producing.
"To be fair, there are hundreds of arrays across different crop species [and] animals but they are not consortia-led public products; they are custom content sets that a small amount of researchers use in internal programs, or one large company has developed using their own internal resources," said Orpin. "There is an enormous number of custom arrays that are being used in many different centers."
This claim was supported by various presentations at last week's meeting. For example, James Cook University's Kyall Zenger discussed using sequencing and custom arrays to create a high-density linkage map for the silver lipped pearl oyster. And Clemson University's Ksenija Gasic discussed the development of genotyping arrays for studying cherry and peach.
Illumina has benefitted from demand for these kinds of chips. While the company does not break out exact revenue numbers by market, Orpin estimated that between 10 and 20 percent of Illumina's annual revenues come specifically from the agriculture sector.
Different Approaches
Depending on the research community, Orpin and Gruia-Gray's predictions for SNP genotyping may or may not be fulfilled. Some groups have already designed arrays or are very close to having a design completed, while others have opted for different platforms for genotyping, such as sequencers or Fluidigm's integrated fluidic circuits for qPCR. And some plant and animal researchers are still working to assemble the genome of their organism of interest, making any resulting genotyping arrays years from being available.
Researchers studying buffalo belong to the first category. Tad Sonstegard, a research geneticist at the US Department of Agriculture's Agricultural Research Service, said that the option of developing a 60,000-SNP chip for genotyping buffalo was discussed at last week's meeting. "We are still in the SNP discovery phase for this project, but we might have a buffalo chip by January 2013 if all goes well," Sonstegard told BioArray News.
Such an array would be "more of an international market product with major efforts in India, Pakistan, Italy … as use of these animals in the US for dairy production is very limited," he noted.
While a buffalo chip may be available within a year's time, Sonstegard questioned the idea that most consortia would eventually adopt genotyping arrays.
"Not all [groups] will use large SNP arrays in a commercial setting, because the value of the ag species may be less than the cost of the genotyping test," said Sonstegard. "In this case other technologies or ways to multiply value of genotypic information to descendants must be devised," he said. "Plants and animals are not very similar, and even the production paradigms and value chains for animals are quite variant," he added.
Another array that may hit the market in the next year is for oat studies. USDA-ARS geneticist Shiaoman Chao said that a SNP array for oat has been three years in the making and was supported in part by US food company General Mills. At PAG, oat researchers discussed the possibility of developing a 1,536-plex array.
"We might bump the number a bit higher, but since we are looking for funds to take us to the next step, we will wait and see what number of SNPs to include in the final array," said Chao.
Chao added that the oat community decided to use Illumina custom arrays because of their "high multiplexing capability."
"Cultivated oat is hexaploid with a large genome size, so we will need a lot of SNPs to cover the entire genome," said Chao. A high-density array will be "useful for trait mapping where we need genome-wide marker distribution," she said. At the same time, breeding applications require fewer SNPs to target specific traits, making other platforms attractive to some users.
"There are labs working with oat that do use [KBioscience's] KASP assay, the Fluidigm system, and melting curves," said Chao. "It's just they have to redesign primers and optimize assays," she said. "Illumina also can work with 48-plex SNPs, the assay is more straightforward, and no primer re-design is needed," Chao noted. "So the decision on which genotyping system to use will depend on the upfront cost for the equipment."
Genotyping by sequencing was also discussed at the oat workshop at PAG. "I won't say we won't be using these techniques ever for oat," Chao said. She noted that these approaches weren't available when the community began generating markers in 2009.
Some researchers have already decided to adopt genotyping by sequencing. USDA-ARS geneticist Lance Cadle-Davidson said that grapevine researchers have used arrays in the past, but that the technology was "too expensive" and "provided too few markers" for performing association studies.
"Our Infinium [Vitis9KSNP] arrays were nominally 9,000-SNP arrays [that] assayed 6,000 loci with good quality, but in a given segregating population, only between 1,000 and 2,000 SNPs were informative," Cadle-Davidson told BioArray News. He said that genotyping by arrays cost about $150 per sample, while by "barcoding and multiplexing 96 to 384 samples in RNA-seq or genotyping-by-sequencing, we get the cost down to $40 per sample for more than 20,000 SNPs," resulting in a "40-fold reduction in per-SNP cost as well as reduced per-sample cost."
Part of the reason the grapevine community has adopted genotyping-by-sequencing has to do with the nature of the fruit crop, which is highly diverse. As grapevine researchers pointed out in a paper last year, the design of a "high-quality genotyping array with millions of SNPs for GWA in the grapevine is, arguably, an impossible task because of the difficulties associated with assaying diversity across such a diverse genus."
The authors recommended that next-generation sequencing “should and will be primarily utilized” for association studies in high-diversity crop species. On the other hand, they argued that customized SNP arrays, such as the Vitis9KSNP chip, will be "valuable for preliminary assessments of germplasm collections and for breeders to verify their material."
Not all research communities are debating the use of genotyping arrays versus genotyping by sequencing. Some are still assembling the genome of their species of interest and lag others in terms of identifying SNPs for future association studies or genomic selection efforts.
In regards to arrays, "there is certainly interest, but we have only just obtained the first cotton reference sequence so are not quite as far along as some of the other communities," said Andrew Paterson, director of the Plant Genome Mapping Laboratory at the University of Georgia. He noted that the polyploidy of cultivated cotton may "necessitate a somewhat different approach than has been taken in other organisms."
Have topics you'd like to see covered in BioArray News? Contact the editor at jpetrone [at] genomeweb [.] com